Abstract: To address the difficulties in integrating numerical models with vibration features in milling surface roughness modeling and simulation, and the insufficient flexibility of the Z-map model for complex surface prediction, a surface roughness simulation method based on point cloud and vibration fusion was proposed using a Tri-Dexel simulation model. Based on theoretical analysis of ball-end milling, a Tri-Dexel point cloud model was established, online vibration signal features of the spindle box were extracted and fused, and a simulation prediction framework for milling surface roughness was constructed. Plane milling experiments were designed to verify the model accuracy from both simulated morphology and numerical values. The results show that the Tri-Dexel model without vibration fusion can predict surface roughness and morphology well, and the prediction accuracy is further improved after fusing vibration signals. In ten groups of bidirectional milling experiments, the average absolute prediction deviation of Sz decreased from 29.3% to 10.5%, that of Sa decreased from 31.0% to 21.9%, and that of Sq decreased from 27.3% to 18.6%. The tool path mode has a considerable influence on surface texture morphology and roughness. Under large process parameters, intensified material deformation and spindle vibration increase the prediction deviation. The proposed method balances simulation efficiency, prediction accuracy, and applicability to complex morphologies, providing a new feasible technical approach for surface roughness control and process parameter optimization.